Deconstructing AFK Arena

Economy design in mobile RPGs

I am currently making my first mobile RPG and deconstructing AFK Arena, one of the references on the market right now. I will update this article while I advance on my economy design journey and focus on different aspects of the progression and monetization.

After I finished developing the core gameplay and the AI of my game, the first question that I asked myself regarding the progression was how the difficulty was going to scale. Many decisions can impact the scaling of the difficulty. How many missions should a player be able to do in one gaming session? How many upgrades should a player need to be strong enough for a new chapter, how often should a player be stuck and have to wait or buy in order to improve?

Why it’s important

Modern mobile RPGs like AFK Arena do not have any real energy system, which means the player could in theory play the game an infinite amount of time until he reaches the end of the content. Because we still need our player to be stuck for various reasons (prevent exhaustion, stretch out the longevity of the game, encourage the player to spend money), the scaling of the difficulty is one of the most important things to get right.

Here is how AFK Arena works. The main game mode is the story, and the player needs to advance through the missions and chapters to unlock rewards. He does not win anything when he fails a mission, and he cannot play previously won missions. The player can try missions as many times as he wants, but he must advance. Because the missions scale in difficulty, his team will need to get stronger if he wants to go forward. This is where the (very simplified) main gameplay loop takes shape: win missions, get rewards, upgrade the team, win more missions. The player goes through this loop until the difficulty outscores the rewards, and the player is stuck and cannot win anymore rewards from playing. This is where the AFK mechanic (that gives resources depending on the time spent between each new session) will subtly invite him to quit the game and come back later.

How to scale the difficulty

But even when you know that your difficulty must scale, you are still left wondering how it needs to scale. I figured the best approach to this matter was to look at the general shape of the power-level curve in AFK Arena. The only data I could find was the minimum player power level required to enter a mission (these values were collected by mrpanafonic and posted on the subreddit of the game). The power level of a player is determined by the power of every hero in his team and power is a value automatically calculated by the game and based on the hero’s statistics. It seems safe to assume that looking at this value is the best way to understand the difficulty of a given mission. Even though this graph only shows the minimum power needed to enter the mission, and not to actually succeed, we can still see that it does scales.

To get more details on the subject, I had to get my own personal data. I decided to play the game every day for about 30 minutes until I felt I had experienced everything the game has to offer and collected enough data. To my surprise, the graph of the actual power of the enemy team in each mission was much uglier than what I expected. It did roughly follow the same growth as the previous graph, but it was filled with spikes and hollows. Analyzing the spikes made me realize that some of the highest points in the curve were immediately followed by some of the lowest. It seems that the intent was to create some kind of variation in the user experience.

One thing that films have become very good at is switching between periods of high action adrenalin and calm. The reason for this is as humans we get saturated with similar stimuli and start to block it out. Knowing when to switch is a real art form, the feeling is also exaggerated by the contrast between the two. […] I think we should be using this kind of technique, not just for emotional things but for difficulty.

David Strachan

Past this reflection, trying to find patterns in the curve was pretty hard. It did seem like every new chapter started with a big climb (probably to compensate for the fact that the player received many rewards from finishing the previous chapter), but that was it. However, while it seemed random, it really was not. As said by Kolen in Dynamic difficulty adjustment for maximized engagement in digital games, one of the main techniques implemented in F2P development is using data to identify undesirable player states like boredom or frustration and adjust the difficulty accordingly. We do not have access to AFK Arena’s internal data, but each new variation in the difficulty could very much be a modification implemented to avoid a specific drop in player retention or to improve conversion at a specific point in the game. While knowing this is very useful, it also means that we cannot just copy this curve, because it was created and modified in reaction to the specific state of the AFK Arena player base.

A “late game” character and his power level

About the late game

If we look at the difficulty scaling on a larger scale (going back to minimum power graph), another thing we notice is the exponential tendency of the curve. This means the increase of power between each level gets bigger and bigger as the player advances. I think this exponential growth is there to solve two main problems such a game could have.

First, because the game is all about getting stronger, the player must get exponentially stronger. By increasing how much more powerful a champion becomes with each level gained, the experience of getting a new level can still feel exciting after a hundred levels. This way, a new level always feels like roughly the same % of power increase. To match the player power growth, the missions need to get exponentially more difficult.

Secondly, the designers need to prevent the player from reaching the end of the content. Depending on time and money spent on the game, different players will consume content at different speeds, and there will always be some player who is really close to the end. If they ever reach it and have nothing else to do, they will surely stop playing the game and try a new one. To avoid this issue, the power gap between missions in the last chapters is extremely big, and the players will be stuck grinding for resources between each level. While it would seem like this system contradicts the design philosophy previously tackled, it seems okay to have « late game players » advance very slowly through the campaign because they will have unlocked a lot of other game modes to enjoy. Furthermore, these players are already extremely invested in the game, so preventing frustration is less important than it is for new players.